Yeast “Make-Accumulate-Consume” Life Strategy Evolved as a Multi-Step Process That Predates the Whole Genome Duplication

When fruits ripen, microbial communities start a fierce competition for the freely available fruit sugars. Three yeast lineages, including baker’s yeast Saccharomyces cerevisiae, have independently developed the metabolic activity to convert simple sugars into ethanol even under fully aerobic conditions. This fermentation capacity, named Crabtree effect, reduces the cell-biomass production but provides in nature a tool to out-compete other microorganisms. Here, we analyzed over forty Saccharomycetaceae yeasts, covering over 200 million years of the evolutionary history, for their carbon metabolism. The experiments were done under strictly controlled and uniform conditions, which has not been done before. We show that the origin of Crabtree effect in Saccharomycetaceae predates the whole genome duplication and became a settled metabolic trait after the split of the S. cerevisiae and Kluyveromyces lineages, and coincided with the origin of modern fruit bearing plants. Our results suggest that ethanol fermentation evolved progressively, involving several successive molecular events that have gradually remodeled the yeast carbon metabolism. While some of the final evolutionary events, like gene duplications of glucose transporters and glycolytic enzymes, have been deduced, the earliest molecular events initiating Crabtree effect are still to be determined.


Introduction
The evolution of plants led to accumulation of larger amounts of mono-and oligo-saccharides, which are now among the favorite substrates for several microbes.Each time when fruits ripen, a fierce competition for the fruit sugars starts, but usually yeasts become the predominant group in these niches.Among the ''winners'' are usually three yeast lineages, including baker's yeast Saccharomyces cerevisiae, which have independently developed the metabolic activity to convert simple sugars into ethanol even under fully aerobic conditions [1,2].During the fermentation the energy for growth is provided by the glycolysis and fermentation pathways, and not by the oxidative respiration pathway.This metabolic activity, called Crabtree effect [3], reduces the production of cell-biomass, but provides a tool, ethanol, to outcompete other microorganisms [1,4,5].Both budding yeast ethanol-producing groups, including S. cerevisiae and Dekkera bruxellensis, can also efficiently catabolize ethanol and therefore their corresponding lifestyle has been named as ''make-accumulate-consume (ethanol)'' strategy [1,4,5].On the other hand, the third Crabtree positive group, including the fission yeast Schizosaccharomyces pombe, can only poorly metabolize ethanol [6].Apart from the three mentioned groups, also a few other single lineages, for example Candida maltosa exhibit a weak Crabtree effect [7].These observations could be interpreted as that (i) Crabtree effect originated early in evolution of Ascomycetes but has been later lost in several lineages, or (ii) the fermentative life style has appeared and been selected simultaneously in several lineages.The onset of yeast genomics [8] has provided a tool to reconstruct several molecular events that have shaped the budding yeasts during their evolutionary history [9].Several molecular events have left a clear finger-print in the modern genomes, while the origin of more complex traits, like the fermentation ability, is often not easy to determine using only a genome analysis approach (figure 1).The whole genome duplication (WGD) [10], which took place app. 100 million years ago (mya), and duplication of the alcohol dehydrogenase encoding gene and genes encoding hexose transporters [4,11] have been proposed as a possible molecular background for development of Crabtree effect and the ''makeaccumulate-consume'' strategy in the S. cerevisiae lineage.On the other hand, the ability of Lachancea yeasts to grow without oxygen [12,13] suggests that the ''invention'' of the ability to grow in the absence of oxygen, took place much earlier, at least 125-150 mya, before the split of the Saccharomyces and Lachancea lineages.
Here, we studied over forty yeast species, which in nature may occupy similar niches and rely on glucose as the ''preferred'' substrate [14], and analyzed their carbon metabolism using uniform experimental conditions during the fully controlled growth in fermenters.Our results can be interpreted as that the ability to produce ethanol under aerobic conditions originated before the WGD and evolved progressively.

Crabtree Positive and Negative Yeasts
The studied species belong to the Saccharomycotina, covering over 200 million years of the yeast evolutionary history [15], and including six WGD genera, Saccharomyces, Kazachstania, Naumovozyma, Nakaseomyces, Tetrapisispora and Vanderwaltozyma, and six non-WGD genera, Zygosaccharomyces, Zygotorulaspora, Torulaspora, Lachancea, Kluyveromyces and Eremothecium [16].The phylogenetic relationship among these yeasts and several molecular events, which has shaped their evolutionary history, and their timing are shown (figure 1).We know that in the modern yeast Saccharomyces cerevisiae at least two of these events, WGD and RGE rewiring, contributed to the observed Crabtree effect, but were during evolution completed at two different time points.In addition, also the expansion of hexose transporters plays a role in efficient Crabtree effect, and this gene duplication occurred in several steps [11].One should note that several alternative phylogenetic trees have been proposed, trying to explain the yeast phylogeny but we followed the one proposed by Kurtzman in ref. 15 (the phylogenetic tree aspect will be discussed further on in the text).
We analyzed several representatives of each genus to get a coherent picture of their carbon metabolism but we excluded the Zygosaccharomyces genus because of its preference for fructose [17] to keep the experimental conditions strictly the same and therefore the results highly comparable (figures 2 and 3).Such a fully controlled experimental approach, covering so many yeast species, has so far not been presented.Crabtree effect results in lower biomass production because a fraction of sugar is converted into ethanol [3].This means that more glucose needs to be consumed to achieve the same yield of cells and this could theoretically result in lower growth rate in Crabtree positive yeasts.In nature, a lower growth rate could have a negative effect in competition with other microbes.In our experiments S. cerevisiae was a reference yeast exhibiting a fully expressed Crabtree effect, with ethanol yield of app.0.39 g per g of glucose and biomass yield of only 0.16 g per g of glucose while Kluyveromyces lactis represented a standard Crabtree negative yeast, which under fully aerobic conditions did not produce any ethanol and its biomass yield was 0.57 g per g glucose.In other words, S. cerevisiae used over 6 g glucose to generate 1 g of biomass, and this biomass produced over 2.5 g of ethanol, while in K. lactis less than 2 g of glucose was needed for 1 g of biomass (table 1; figures 3 and S1).However, even if they used a different fraction of glucose for the generation of new biomass, the growth rate of both yeasts was very similar, 0.287 for S. cerevisiae and 0.298 for K. lactis (table 1 and figure 4).Regarding Crabtree positive yeasts, in a majority of strains, including S. cerevisiae, the glucose was completely depleted at the point where  Kurtzman and Robnett (2003) [16], covering twelve genera of Saccharomycetaceae and all employed species.Note that alternative models to explain the phylogenetic relationship between the Lachancea, Kluyveromyces and Eremothecium genera have been proposed [23] but here we follow the tree in ref. 15.Several evolutionary events, which are relevant for the modern traits, are shown.Note that the relative timing of some events, especially those which left a clear finger-print in the modern genomes (green arrows) is relatively precise, such as WGD [10], the horizontal transfer of a bacterial »anaerobic« DHODase (encoded by URA1) [19], complete rewiring of the respiration related promoters (RGE stands for Rapid Growth Elements) [3], and the loss of respiratory Complex I [9], while the timing of more complex traits (red arrows), such as the capability for anaerobic growth [12,13] and petite positivity [13], might be less precise.doi:10.1371/journal.pone.0068734.g001ethanol concentration reached the maximum and started to be utilized as a carbon source.This gives a very sharp border between the fermentative and respiratory metabolism.In a few yeasts, like E. coryli and Z. rouxii, the ethanol maximum was reached when some glucose was still present.This means that there was a time period of mixed metabolism, fermentative and respiratory (table S1).Anyhow, we attempted to treat all species and the obtained fermentation results in a comparable way (see also Materials and Methods).

The Origin of Crabtree Effect
The analyzed yeasts converted glucose into biomass, ethanol and CO 2 , and only traceable amounts of other products, such as acetate and glycerol (table S1).When we analyzed how many grams of glucose were necessary to obtain 1 g of biomass, and what the yield of ethanol was (table 1; figures 2 and 3), we could arbitrarily divide the studied species into three different groups: (i) a group which needed over 5.5 g of glucose for 1 g of biomass (or ,0.18 g biomass generated from 1 g glucose) and this then produced over 2 g of ethanol (or .0.33 g ethanol generated from 1 g glucose), and included a great majority of WGD yeasts (genera Saccharomyces, Kazachstania, Naumovozyma and Nakaseomyces (but not N. bacillisporus and N. castellii), Tetrapisispora (but not T. iriomotensis) and Vanderwaltozyma); (ii) a group which needed between 3-5 g of glucose for 1 g biomass (or 0.20-0.33g of biomass from 1 g glucose), and included Zygotorulaspora, Torulaspora, Lachancea yeasts (which are non-WGD) and the lower branches of WGD yeasts (see the Tetrapisispora and Nakaseomyces strains mentioned above); (iii) a group which converted a majority of glucose into biomass, so that less than 3 g glucose was needed for 1 g of biomass (or .0.33 g of sinecaudum (F).The graphs show time dependence of yeast glucose consumption, and appearance of the fermentation products and biomass.The ethanol and biomass yields vary enormously among the six shown species, as well as among other studied yeasts (table 1), and are related to the phylogenetic position of each studied yeast species (see also figure 1).doi:10.1371/journal.pone.0068734.g002biomass from 1 g glucose) and almost no ethanol or less than 0.15 g from 1 g of glucose was generated.On average, group 1 produced more ethanol per 1 g glucose than group 2. We then performed a statistical analysis of the results shown in figure 3, but for this purpose we build new groups based on their phylogenetic position.
The yeast genera were grouped into four groups, based on their phylogenetic relationship and some of the evolutionary steps shown in figure 1, like WGD, the settlement of the RGE rewiring and the origin of anaerobic growth.Group 1 included all tested species belonging to Eremothecium and Kluyveromyces, group 2 all strains of the Lachancea, Torulaspora and Zygotorulaspora genera, group 3 contained the Vandervaltozyma and Tetrapisispora, and group 4 all species belonging to the Saccharomyces, Kazachstania, Naumovozyma and Nakaseomyces genera.When biomass and ethanol yields were compared for the four groups a large gap, regarding the average values for ethanol yield and biomass yield between group 1 and the rest was found (figures 5 A and B; tables 1 and 2).These differences were highly significant, both tested with regular ANOVA but also when a non-parametric test was used (table 2).Table 2 also shows pairwise t-tests showing significant differences between group 2 and 4, while group 3 overlapped with group 2 as well as 4.
In general, the ethanol and biomass yields of each species correspond to its phylogenetic position, or in other words, closely related species exhibit similar traits.This pattern can be interpreted as that a clear Crabtree effect originated just after the split of the S. cerevisiae lineage (including all WGD yeasts and Zygotorulaspora, Torulaspora, Lachancea lineages) from the Kluyveromyces lineage.However, the observed Crabtree effect was much more pronounced in a majority of WGD yeasts, than in the ethanol producing non-WGD species, suggesting a gradual or at least a two-step ''invention''.
Carbon metabolism in the ''lower'' branches of Saccharomycetaceae yeasts, belonging to the modern Kluyveromyces and Eremothecium species, is similar to other Saccharomycotina yeasts, like Candida albicans, Yarrowia lipolytica and Pichia pastoris, which are Crabtree negative yeasts [18] (table 1 and figure 3), confirming that this could be the original property of the Saccharomycetaceae yeasts.Therefore, the origin of the ''make-accumulate-consume'' strategy could take place within the time interval spanning the origin of the ability to grow under anaerobic conditions [12], the URA1 horizontal transfer [19] and loss of respiratory chain Complex I (figure 1).On the other hand, the second step, leading towards even a more pronounced Crabtree effect, occurred relatively close to the WGD event [10], the settlement of rewiring of the promoters involved in the respiratory part of the carbon metabolism [20], and the settlement of the petite-positive character [13] (figure 1).It appears that Vanderwaltozyma polyspora has not yet completely rewired its promoters regarding the RGEelement, and that this process was in yeast first completed after the separation of the Vanderwaltozyma and Saccharomyces lineages [21].Thus we consider this evolutionary event to be first completed/ settled after the separation of these two lineages (figure 1).Similarly, the petite positive character can be found in all Saccharomyces, Kazachstania, Naumovozyma and Nakaseomyces species, while Tetrapisispora and Vanderwaltozyma contain petite-positive and -negative species [13], thus the trait is first settled after the separation of the Vanderwaltozyma and Saccharomyces lineages.Apparently, all three characters, the complete promoter rewiring [20], settled petite-positivity [13] and strong Crabtree effect (figure 3), occurred slightly after the WGD event.The first branches after the WGD event, Tetrapisispora and Vanderwaltozyma, represent a kind of intermediate lineages where the three traits are still in transition.

Improved Consumption of Glucose
Another interesting aspect is the relationship between Crabtree effect and the yeast growth rate and glucose consumption rate (g glucose/g biomass/time).These parameters were calculated in all species when they grew exponentially on glucose solely (before they started using any accumulated ethanol).In our experiments, Crabtree negative group, Kluyveromyces and Eremothecium, exhibits a moderate glucose consumption rate, under 0.70 g glucose/g biomass/hour (figure 4).This rate was almost doubled in the Zygotorulaspora, Torulaspora, Lachancea yeasts, and almost tripled, over 1.75 g glucose/g biomass/hour, in some WGD genera, Saccharomyces and Kazachstania, while Naumovozyma, Nakaseomyces, Tetrapisispora and Vanderwaltozyma were rather similar to the ''intermediate'' group (Zygotorulaspora, Torulaspora and Lachancea yeasts).
The four groups were also compared with respect to, glucose consumption rate and growth rate (figures 5 C and D; tables 1 and 2).Glucose consumption rate showed a similar pattern to ethanol and biomass yield above, i.e. there was a large gap, regarding the Figure 3. Yeast ethanol and biomass yield.Different yeast species were studied for their carbon metabolism: ethanol yield as g of ethanol per g of glucose (red), and biomass yield as g of biomass per g of glucose (blue).Detailed results and biological replicates are shown in table 1 but hereby either a single measurement or an average of two replicates are presented.The yeast species are presented starting with the Saccharomyces genus at the top and then following a decreasing phylogenetic relationship, following figure 1.The species related the least to S. cerevisiae are at the bottom, and the gap divides the Saccharomycotina and non-Saccharomycotina yeasts.The four groups of yeasts (1, 2, 3 and 4) used in the statistical analysis are shown.In general, the ethanol yield gradually drops and the biomass yield gradually increases with the genetic distance from the Saccharomyces yeasts (see also figure 5 A and B).doi:10.1371/journal.pone.0068734.g003 average values for glucose consumption rate between group 1 and the rest, differences being highly significant.There were also significant differences between group 2 and 4, while group 3 overlapped with group 2 as well as 4. On the other hand, there were no significant differences between the four groups regarding the growth rate (figure 5 D and table 2).When the pronounced Crabtree positive yeasts were considered separately it was found that Saccharomyces and Kazachstania exhibited a very similar growth rate of approximately 0.25-0.35/hcomparing with the clear Crabtree negative yeasts, for example the Kluyveromyces species (figure 4), even if these convert a majority of sugars into biomass.In other words, Crabtree positive yeasts exhibited a similar growth rate capacity even if they ''wasted'' a part of the carbon source for ethanol production (see table 2).When all species were analyzed there was a clear correlation between biomass yield and glucose consumption rate (figure S4 and table 2).On the other hand, no significant correlation between growth rate and any of the other variables could be found.This is of course in consequence with the absence of significant differences in growth rate between groups.Again, a progressive evolution towards improved consumption can be noticed and it likely occurred in at least two steps.These observations coincide with the gradual gene duplications of hexose transporters [11].

Evolutionary Timing of Crabtree Effect
Crabtree effect could originate early in the evolution of Ascomycetes but has been later lost in several lineages, or the .Yeast glucose consumption, ethanol production and growth rates.Different yeast species were studied for their carbon metabolism and the results are shown as: specific glucose consumption rate (g/g h -1 amount of consumed glucose by 1 g of biomass and multiplied with the corresponding specific growth rate, shown in blue), specific ethanol production rate (g/g h -1 ethanol produced by 1 g biomass and multiplied with the corresponding specific growth rate, shown in red), and specific growth rate (per hour, shown in green).Detailed results and biological replicates are shown in table 1 but hereby either a single measurement or an average of two replicates are presented.The yeast species are presented starting with the Saccharomyces genus at the top and then following a decreasing phylogenetic relationship, according to figure 1.The species related the least to S. cerevisiae are at the bottom, and the gap divides the Saccharomycotina and non-Saccharomycotina yeasts.The four groups of yeasts (1, 2, 3 and 4) used in the statistical analysis are shown (see also figure 3).Specific glucose consumption rate decreases with the phylogenetic distance from the Saccharomyces genus, indicating that Crabtree negative yeasts have only a moderate rate, while Crabtree positive have faster glucose consumption rate.On the other hand, the growth rate (in green) is very similar among all Saccharomycotina yeasts (see also figure 5 C and D).doi:10.1371/journal.pone.0068734.g004 fermentative life style has originated independently in several lineages.The previously obtained results on a very few Saccharomycetaceae yeasts could also be interpreted as that Crabtree effect might have originated independently within this monophyletic group [13].However, in Saccharomycetaceae a few independent steps, including WGD, RGE-element rewiring and several independent gene duplications of glucose transporters, which took place at different time points, have been deduced.This step-wise event strategy, apparently strengthened Crabtree effect in the Saccharomyces lineage, and suggests that the Saccharomycetaceae progenitor was a Crabtree negative yeast.The origin of modern plants with fruits, more than 125 mya [22], brought to microbial communities a new larger and increasingly abundant source of food based on simple sugars.However, ancient yeasts could hardly produce the same amount of new biomass as bacteria during the same time interval.Slower growth rate could in principle be counter-acted by production of a compound, which could inhibit the growth rate of bacteria, like ethanol and acetate.Here we demonstrate that the origin of the Crabtree effect in the Saccharomyces lineage took place at an earlier period (figures 1 and 3), 125-150 mya, then suggested before, and it possibly coincided with the independent origin of the Crabtree effect in the Sch.pombe lineage and in the D. bruxellensis lineage [1,2] and occurred at the similar time point as the origin of the first modern fruits [22].While we know the later events, which strengthened the trait, what could be the initial molecular mechanisms, which promoted the evolution of this new lifestyle and rewiring of the carbon metabolism?

Concluding Remarks
The tremendous change in aerobic/anaerobic properties and carbon metabolism took place just after the split of the Kluyveromyces lineage and the lineages leading to Lachancea/Zygotorulaspora/ Figure 5. Statistical analysis of the fermentation parameters.Yeasts have been grouped into the following groups: Group 1 included all tested species belonging to Eremothecium and Kluyveromyces, group 2 all strains of the Lachancea, Torulaspora and Zygotorulaspora genera, group 3 contain the Vandervaltozyma and Tetrapisispora, and group 4 all species belonging to the Saccharomyces, Kazachstania, Naumovozyma and Nakaseomyces genera.The group mean values on four parameters: ethanol yield (A), biomass yield (B), glucose consumption rate (C) and growth rate (D) are illustrated.All error bars in the figure cover a 95% confidence interval for each group.Group 3 appears to be an intermediate between group 2 and 4 for all parameters except growth rate.Hence, no significant difference in growth rates among groups can be observed.These results are also supported by statistical analysis of variance and pairwise t-test (table 2).doi:10.1371/journal.pone.0068734.g005This table summarizes the results from the statistical analysis conducted on our results.Tests between two parameters (under x-and y-column) are considered to be significant at a significance level (alpha = 5%) or p-value lower than 0.05.ANOVA and Kruskal-Wallis test amongst the four groups reveal significant differences in all parameters (except for growth rate).Hence, there is no significant difference in growth rate among groups 1, 2, 3 and 4. The groups were defined in the main text as well as in figure 5. Pairwise t-tests performed separately on all combinations of groups, on all variables (except growth rate) reveal significant differences between all group combinations (except between groups 2-3, and groups 3-4).Once more, t-test failed to detect any significant differences in growth rates between groups 1, 2, 3 and 4, and the results indicate that group 3 can be seen as an intermediate between group 2 and 4 (see also figure 5).A highly significant correlation between glucose consumption rate and biomass yield is seen on both parametric and non-parametric tests.Thus, our data support a linear model that explains 49% of the variation (See also table S3 for comparison within groups).Furthermore a significant correlation between glucose consumption rate and growth rate, and no significant correlation between growth rate and ethanol yield can be seen with both parametric and non-parametric tests.doi:10.1371/journal.pone.0068734.t002 Torulaspora/WGD lineages (table 1; figures 2 and 3).However, this interpretation is based on the phylogenetic relationship [16] that the Saccharomyces-Lachancea clade is monophyletic regarding Kluyveromyces and Eremothecium.However, several authors (reviewed in [23]) propose that the correct topology is that Lachancea, Kluyveromyces and Eremothecium form a monophyletic group that is sister to the WGD yeasts.The yeast phylogeny is still controversial but if the later topology is true, it means that the progenitor of all three genera, Lachancea, Kluyveromyces and Eremothecium, was anaerobic and Crabtree positive.This would also claim that the origin of these traits is much older, probably close to the loss of Respiratory complex I.One could then further interpret our results as that the two genera, Kluyveromyces and Eremothecium, later lost the two traits.What can one deduce when comparing the published results on physiology of the Kluyveromyces and Lachancea/WGD species?S. cerevisiae and L. kluyverii have a Rox1p-mediated system responding to oxygen limiting conditions [24], which regulates gene expression through specific promoter motifs present in hundreds of genes.Similarly, a Mig1p-mediated glucose repression system that is at least partially down-regulating the respiration associated genes, operating in Saccharomyces but not in Kluyveromyces yeasts, is one of the mechanisms involved in the switch between the fermentative and respiratory mode [25,26].The enlargement of a global regulatory system, through ''spreading'' of the regulatory motifs into new genes, to be controlled for example by Rox1p or Mig1p, could promote colonization of progressively anaerobic niches or promote the ability to more efficiently poison competing bacteria.The URA1 horizontal transfer [19,27] and establishment of efficient sterol uptake system [28] also strengthened the Crabtree effect and facultative anaerobiosis phenotypes.Several gene duplications gained during further evolutionary steps increased the carbon flow through glycolysis [29], optimized the conversion between acetaldehyde and ethanol [4], and elevated the sugar uptake.It can be concluded that the ancient environment consisted of lucrative new niches, with an excess of simple sugars originating from the modern plant fruits, which promoted independent evolution of carbon metabolism remodeling in at least three yeast lineages.These could either employ similar or different evolutionary pathways [1] to achieve very similar modern traits.

Yeast Strains
The yeast species, which were characterized for their respirofermentative properties in this study, belong to the Ascomycota phylum, sub-phylum Saccharomycotina, clade Saccharomycetaceae.The species are presented according to the phylogenetic analysis presented by Kurtzman [16] In addition, some species belonging to other Ascomycota clades were studied: Debaromyces vanrijiae CBS 3024 (NRRL Y-7430), Dekkera bruxellensis CBS 2796, Pichia philogaea CBS 6696 (NRRL Y-7813), Pichia pastoris Y1294, Schizosaccharomyces pombe Eg282 and represent a control group.All batch cultivations were verified by sequencing of the inoculum and the final culture (table S1).
Regarding Zygosaccharomyces bisporus CBS 702 (NRRL Y-12626) and Zygosaccharomyces rouxii CBS 732 (NRRL Y-229), this genus is fructophilic [17] and was therefore not included in the comparisons shown in figures 3 and 4.Moreover, Tetrapisispora blattae CBS 6284 (NRRL Y-10934) was isolated from the gut of Blatta orientalis (cockroach) and was therefore also excluded from the comparisons in figures 3 and 4.However, the results on these three strains can be found in supplementary information.

Sequencing
All batch cultivations were verified by sequencing of the rDNA R26 region, amplified by NL-1 (59-GCATATCAATAAGCG-GAGGAAAAG-39) and NL-4 (59-GGTCCGTGTTTCAA-GACGG-39) primers.Samples from the end of each batchcultivation were streaked out on YPD-agar plates (containing 2% D-Glucose) and incubated at 25uC for 2-3 days.Template for colony PCR was prepared by picking small amounts of cells from a single colony with a sterile pipette tip.Cells were resuspended in 20 ml of distilled water and 1 ml of the suspension was used in PCR reactions containing: 1.25 U Taq DNA polymerase (Fermentas), 1X Taq buffer with (NH 4 ) 2 SO 4 (Fermentas), 2.5 mM MgCl 2 (Fermentas), 2 mM dNTP mix (Promega), 0.4 mM of each primer (Eurofins MWG operon) and sterile MQ-H 2 O were added to a final volume of 50 ml.The PCR-reactions were run with the program: 94uC -4 min, 36 * (94uC -30 sec, 52uC -30 sec, 72uC -1 min), 72uC -7 min, 4uC '.PCR-product quality was verified with gel electrophoresis; 1% agarose gel and 1X TBE buffer as a mobile phase.Nanodrop was used for quantification and Eurofins MWG Operon provided the sequencing service.A normal nucleotide blast (Blastn) of NL-1 derived sequence was performed to finalize the verification of species used in the experiments.

Batch Cultivations
Aerobic batch cultivations were performed in Multifors (INFORS HT) bioreactors with a working volume of either 0.5 or 1 liter.A majority of all batch cultivations were conducted in duplicate at 25uC, with airflow of 1 vvm.The stirrer speed was on cascade mode, automatically varying from 200 to 1200 rpm to maintain a dissolved oxygen concentration above 30%, which was monitored with an InPro 6800S sensor (Mettler Toledo).The pH was maintained at 5 (60.5 units) by KOH (2 M) and H 2 SO 4 (1 M), monitored with a 405-DPAS-SC-K8S/225 (Mettler Toledo) sensor.For the calculation of CO 2 production and O 2 consumption, which were used to estimate respiration contra fermentation and Carbon-balancing calculations, gas analyzers BC-CO 2 and BCP-O 2 (BlueSens) were used to determine the CO 2 and O 2 levels in the gas outflow.All overnight pre-inoculum were cultivated and washed before inoculation in defined synthetic minimal medium for aerobic conditions, used in the bioreactors and prepared as reported [30].Pre-inoculums were approx.500fold diluted at the inoculation step, resulting in approximately 500 times higher biomass concentration in the bioreactors, in the end of each experiment.Furthermore, 2.002 g/l Kaiser Synthetic Complete supplement had to be added to batch cultivations of Eremothecium coryli, and 150 mg/l histidine to batch cultivations of Tetrapisispora blattae due to the auxotrophic nature of the species.In all experiments the only carbon source to be utilized by the studied species was 2% glucose.

Yeast Central Carbon Metabolism
The determination of glucose, ethanol, acetate and glycerol in the supernatant were conducted with a HPLC 1200 series (Agilent) equipped with a 300*7.7 mm Aminex HPX-87H Column (BioRad).The mobile phase used was 5 mM H 2 SO 4 set to a flow rate of 0.6 ml/min through a column temperature of 60uC and a RID temperature of 55uC.Growth kinetics was monitored by two methodologies, dry weight (DW) and optical density (OD 600 ) measurements (tables 1 and S1).The maximum growth rates were calculated in the exponential growth phase on glucose solely, before the consumption of any accumulated ethanol (figure S1).For the determination of DW, glass microfiber filters GF/A (Whatman) were weighted before and after filtering of the washed culture samples (these were dried at 70uC for 1 day).
Yields of products (biomass, ethanol, acetate and glycerol) relative to consumed substrate (glucose) are presented in the unit g/g and were calculated by dividing the amount of product (at the maximum of ethanol concentration for Crabtree positive yeast, or when glucose was depleted for Crabtree negative yeast) with the amount of substrate consumed (tables 1 and S1).
The consumption rates of substrate (glucose) and production rate of products (biomass, ethanol, acetate and glycerol) are presented in the unit g/gDW,h (gram substrate or product per gram biomass per hour).These rates were calculated between the time points (the same as for the determination of the maximum growth rates) that span the exponential growth phase on glucose solely (figure S1), by dividing the amount of consumed glucose or formed product with the amount of produced biomass, and multiplied with the corresponding specific growth rate.Thus, by dividing the yields of products relative to the consumed glucose with the yield of biomass relative to the consumed glucose (table S2) and multiply with the corresponding specific growth rate (table 1), the consumption and production rates were obtained.
The respiration ratios were calculated by dividing the total amount of CO 2 produced with total amount O 2 consumed, during growth on glucose solely, in the unit mole/mole (figure S2).To verify the quality of each experiment, carbon balance was calculated by taking the ratio of formed products in C-mole and consumed substrate in C-mole (figure S3).

Statistical Analysis
Statistical tests and correlation analysis were performed in R (2.15.2) and summarized in table 2 and S3.Analysis of variance (ANOVA) and non-parametric (Kruskal-Wallis) tests were applied on four parameters (biomass yield, ethanol yield, glucose consumption rate and growth rate) to assess any significant difference among groups (shown in figures 3, 4 and 5).Subsequent pairwise t-tests (Welch two sample t-test) on the group-means of the same four parameters (mentioned above) were then performed between all possible combinations of groups.More specific information on the statistical analysis can be found in Supplementary Material S1.

Supporting Information
Figure S1 Yeast growth profiles.All characterized yeast species and their growth profiles are shown, both in natural and logarithmic scale.Specific rates were determined from dry weight (DW) and optical density (OD 600 ) (illustrated in figure 4, and summarized in table 1 and S1).Substrate (glucose) and metabolite (pyruvate, succinate, lactate, glycerol, acetate and ethanol) concentrations were monitored during growth and were used for yield, production/consumption rates calculation to quantify the Crabtree effect for each species (illustrated in figures 3, 4, S2 and summarized in tables 1 and S1).(PDF) Figure S2 Respiration ratio.The respiration ratios for different species illustrates the activity of alcohol fermentation pathway as compared to respiratory pathway and was calculated by dividing the total amount of CO 2 produced (blue bar) with total amount O 2 consumed (red bar) in the unit mole/mole (see also table S1).The yeast species are presented starting with the Saccharomyces genus at the top and then following a decreasing phylogenetic relationship, following figure 1.The species related the least to Saccharomyces cerevisiae are at the bottom, and the gap divides the Saccharomycotina and non-Saccharomycotina yeasts.In general, the respiration ratio gradually drops with the phylogenetic distance from the Saccharomyces yeasts.(PDF) Figure S3 Carbon balance.To verify the quality of each experiment, carbon balance was calculated by taking the yield ratio between measured products in C-mole and consumed substrate in C-mole, see also tables 1 and S1 for data on products yield and substrate consumption.(PDF) Figure S4 Correlation between biomass yield and glucose consumption rate.A significant correlation can be observed between the determined biomass yield and glucose consumption rate for each species (table 1).The rate-yield trade-off is a known phenomenon, which has been observed previously and hypothesized to act as an evolutionary constraint [31].(PDF) Table S1 Central carbon metabolism of characterized yeast species in this study.Different yeast species were studied for their carbon metabolism and the results for all analyzed metabolites, not illustrated in figures 3 and 4, are summarized in this table: Yield of products (glycerol, acetate, pyruvate, succinate and lactate) relative to consumed substrate (glucose) is presented in the unit g/g (gram product per gram substrate) and is calculated by dividing the amount of product (at the maximum of ethanol concentration for Crabtree positive yeasts, or when glucose was depleted for Crabtree negative yeasts) with the amount of substrate consumed.Yield of CO 2 and consumed O 2 relative to consumed substrate is presented in the unit mole/mole (mole CO 2 or O 2 per mole glucose) and is determined in the same time interval as for other yield calculations.Characterized species names, accession number to their best blast hit and Y-collection numbers are mentioned in separate columns.(XLSX) Table S2 Central carbon metabolism under exponential growth phase on glucose solely.Yields of all determined products (formed metabolites) relative to consumed substrate (glucose) are presented in the unit g/g (gram product per gram substrate) and is calculated by dividing the amount of product formed during the exponential growth phase on glucose solely, before the consumption of any accumulated ethanol.This may not correspond to the maximum of ethanol concentration for Crabtree positive yeasts, or when glucose was completely depleted for Crabtree negative yeasts (as compared to the yield data summarized in tables 1 and S1).(XLSX) Table S3 Correlation between glucose consumption rate and biomass yield within groups.No significant correlation between growth rate and ethanol yield can be observed within phylogenetic groups on both parametric and non-parametric tests.(XLSX)

Figure 1 .
Figure 1.Phylogenetic relationship among yeast.A schematic phylogenetic relationship, based on the phylogenetic tree fromKurtzman and Robnett (2003) [16], covering twelve genera of Saccharomycetaceae and all employed species.Note that alternative models to explain the phylogenetic relationship between the Lachancea, Kluyveromyces and Eremothecium genera have been proposed[23] but here we follow the tree in ref.15.Several evolutionary events, which are relevant for the modern traits, are shown.Note that the relative timing of some events, especially those which left a clear finger-print in the modern genomes (green arrows) is relatively precise, such as WGD[10], the horizontal transfer of a bacterial »anaerobic« DHODase (encoded by URA1)[19], complete rewiring of the respiration related promoters (RGE stands for Rapid Growth Elements)[3], and the loss of respiratory Complex I[9], while the timing of more complex traits (red arrows), such as the capability for anaerobic growth[12,13] and petite positivity[13], might be less precise.doi:10.1371/journal.pone.0068734.g001

Figure 2 .
Figure 2. Yeast growth profiles.A few examples of a representative batch culture experiment showing different capacity to produce ethanol and biomass in the presence of excess glucose and oxygen: Sac.eubayanus (A), Kaz.exiguus (B), Zto.mrakii (C), Lac.waltii (D), Klu.wickerhamii (E) and Ere.sinecaudum (F).The graphs show time dependence of yeast glucose consumption, and appearance of the fermentation products and biomass.The ethanol and biomass yields vary enormously among the six shown species, as well as among other studied yeasts (table1), and are related to the phylogenetic position of each studied yeast species (see also figure1).doi:10.1371/journal.pone.0068734.g002

Figure 4
Figure 4. Yeast glucose consumption, ethanol production and growth rates.Different yeast species were studied for their carbon metabolism and the results are shown as: specific glucose consumption rate (g/g h -1 amount of consumed glucose by 1 g of biomass and multiplied with the corresponding specific growth rate, shown in blue), specific ethanol production rate (g/g h -1 ethanol produced by 1 g biomass and multiplied with the corresponding specific growth rate, shown in red), and specific growth rate (per hour, shown in green).Detailed results and biological replicates are shown in table 1 but hereby either a single measurement or an average of two replicates are presented.The yeast species are presented starting with the Saccharomyces genus at the top and then following a decreasing phylogenetic relationship, according to figure1.The species related the least to S. cerevisiae are at the bottom, and the gap divides the Saccharomycotina and non-Saccharomycotina yeasts.The four groups of yeasts (1, 2, 3 and 4) used in the statistical analysis are shown (see also figure3).Specific glucose consumption rate decreases with the phylogenetic distance from the Saccharomyces genus, indicating that Crabtree negative yeasts have only a moderate rate, while Crabtree positive have faster glucose consumption rate.On the other hand, the growth rate (in green) is very similar among all Saccharomycotina yeasts (see also figure5 C and D).doi:10.1371/journal.pone.0068734.g004

Table 1 .
Central carbon metabolism of characterized yeast species in this study.
*Growth rate determined from DW measurements Different yeast species were studied for their carbon metabolism and the results illustrated in figures 3 and 4 are summarized in this table.Yield of products (biomass and ethanol) relative to consumed substrate (glucose) is presented in the unit g/g (gram product per gram substrate) and is calculated by dividing the amount of product (at the maximum of ethanol concentration for Crabtree positive yeasts, or when glucose was depleted for Crabtree negative yeasts) with the amount of substrate consumed.The consumption rates of substrate and production rate of products is presented in the unit g/g,h (gram substrate or product per gram biomass per hour), and is calculated during the exponential phase of growth by dividing the amount of consumed glucose or produced product with the amount of produced biomass and multiplied with the corresponding specific growth rate.Characterized species names and different collections (Y, CBS and other) that provide them are mentioned in separate columns.doi:10.1371/journal.pone.0068734.t001

Table 2 .
Statistical test and correlation analysis.